Behavioral Analysis of Web Services for User Categorization based on Stochastic Models

Abstract

newline The need for web applications is on the increasing side as a result of the growing internet newlineusers. Once a service request is made to an application, the functionality happening behind is newlineexplained in terms of web services. Though the technologies and web frameworks have been newlineproviding support in terms of programming languages, the users still look for better responses. newlineBut certain responses are hard to diagnose the reason in web applications. Hence, it is impor tant to diagnose the web services behavior in different scenarios and states which will help in newlineimproving the web service response. This is the main idea of this research. newlineThis research has focused on understanding the response code with a feature set genera tion. As a preliminary analysis, the different web services under different categories including newlinethe WSDream data set was executed and features relevant to all states were gathered. As the newlineWSDream Dataset had the limitation of producer a full-fledged application named e-Job (Job newlinePortal) was created with web services customized for users. The application was then deployed newlinein an in-house environment for students to access and their web user logs were generated. It newlineprovided a way for generating a new feature set for web service user logs based on states. newlineThe research next focuses on proposing a new Finite State Machine Model that could allow newlinefor understanding the transitions among the states. A real-time mapping to user experiences in newlineaccessing web services was analyzed and the performance is evaluated in terms of Total Re sponse Time. The novelty was in proposing a Finite State Machine Model for Web Service user newlineinteractions. newlineThe research then was extended to understanding how the classification of web service user newlinelogs could be done. As there was no benchmark classification for the web service user logs newlineK-Means clustering was used for classification. Then algorithms like KNN, Fuzzy C-Means newlinewere used for predicting the test data. As the web service user log data is a time-dependent.